strands-agents / box
Run AI agents in a sandbox that restricts what they can execute, read, write, and reach on the network. Box combines OS isolation with default-deny Dogwood policies and credential injection that keeps secrets outside the agent. Written in Rust. Supports macOS on Apple silicon and Linux.
Why this job: topics: security, sandbox, ai-safety
What it does
- File, program, and network restrictions. The agent's sandbox limits its
- Semantic and temporal policies. Dogwood rules can depend on the requested
- Policy across code and tools. Strands Shell, Monty for Python, the egress
- Request and tool-call checks. The egress gateway checks connections and
- Credential injection. The gateway authenticates permitted requests with
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Active
- Last commit yesterday
- Stars
- 512
- +20 this week
- Licence
- Apache-2.0
- Allows business use
- Downloads for
- 2 systems
- Mac, Linux
- Latest version
- v0.2.1
- Released 4 days ago
- Built with
- Rust
- 95% of the code
Is it looked after?
- Contributors
- 15
- Last 90 days
- Issues
- 24% closed
- 64 still open
- Pull requests
- 159 merged
- 52 waiting
- Releases
- 3
- 3 in the last year
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| eugene1g / agent-safehouse | 2,146 | 2 days ago | Apache-2.0 | No downloads |
| dredozubov / hazmat | 180 | 13 days ago | MIT | Mac |
| provos / ironcurtain | 657 | yesterday | Apache-2.0 | No downloads |
More: download files, languages, topics and project record
Download files, v0.2.1
- Mac1 file, 17 downloads
- Linux1 file, 66 downloads
Languages
- Rust 95%
- Shell 2%
- Python 2%
- C <1%
Topics
- agentic-ai
- agents
- ai
- ai-agents
- ai-safety
- containment
- egress
- harness
- linux
- llm
- macos
- mcp
Project record
- JobAI security
- StatusFeatured
- Commits159
- Forks43
- Watching1
- Stars in 28 days+120
- CreatedOct 2, 2026
- First seen by usNov 7, 2026
- CheckedNov 10, 2026